Survey Methodology EPID 626
Exercise # 3

Due 03/22/00

Please do your own work; all provisions of the honor code should be adhered to.

Eliciting Sensitive Information - The application of computer assisted interviews to sexually transmitted disease research

Scenario: Ascertaining information that may be considered sensitive by some persons is difficult. Recently computer-assisted/self-administered (CASI) interviews have been used to promote higher validity of answers to these questions. We conducted a study to examine if young women would be more likely to answer yes to socially undesirable behaviors on CASI and less likely to answer yes on CASI to socially acceptable behaviors. We enrolled women 14 years of age or older, who had a diagnosis of Chlamydia trachomatis and were a part of a larger cohort study. The women were randomized to receive either the CASI first or a face-to-face interview (FTFI) first. Immediately upon completion of the randomized modality, the second type of interview was conducted.

Coding Guide: Click here to download compressed file (VCASI)
 

Variable Description and Code
PTID ID number
AGE Continuous, 0=missing
RACE 1=Black, 2=White, 3=American Indian, 4=Asian Pacific Islander, 5=Other, 9=Unknown
COMF Comfort with computers 1=Very comfortable/comfortable,
2=Somewhat/not at all comfortable
SEXPART More than two sex partners in last 60 days 1=Yes, 2=No,
9=Unknown
VSEXPART Same question as SEXPART asked on computer
NSEXPART New sex partners in last 60 days 1=Yes, 2=No, 9=Unknown
VNEWSEX Same questions as NSEXPART asked on computer
PRIORGC Ever had a prior gonnorrhea infection 1=Yes, 2=No, 3=Don't know
VPRIORGC Same question as Priorgc asked on the computer
DOUCH Have you ever douched in the last month 1=Yes, 2=No, 3=Unknown
VDOUCH Same question as douch asked on the computer
CONDUSE Used a condom last vaginal sex 1=Yes 2=No 9=Unknown
VCONDUSE Same as CONDUSE asked on computer
CONTIME Used condoms more than half the time 1=Yes, 2=No
VCONTIME Same question as contime asked on the computer
PREGNOW Are you currently pregnant 1=Yes, 2=No, 9=Unknown
VPREG Same question as pregnow asked on the computer
SYMPT Are you symptomatic at this visit 1=Yes, 2=No, 9=Unknown
VSYMPT Same question a sympt asked on the computer
RAN Randomization order 0=FTFI first, 1=CASI first
  1. Please fill in this table

Table 1. Assessment of socially desirable, undesirable and neutral questions by interview modality (N=    )
 

  V-CASI
% responding yes		 FTFI
% responding yes		 Crude OR
V-CASI/FTFI		 Kappa Score
Socially undesirable        
> 2 sex Partners        
 New Sex Partners         
 History of gonorrhea        
 Douching        
Socially desirable        
Used Condom last vaginal intercourse        
Used Condom more than half of the time        
Neutral        
Pregnancy        
Symptomatic at this visit        

For a definition of Kappa statistics CLICK HERE

  1. Is there any trend for women answering yes to socially undesirable behaviors on CASI and yes more often for socially desirable behaviors on FTFI? If so, what? If not, why not?

 

  1. Which questions have good reliability (i.e. Kappa score > .75), adequate reliability (i.e. Kappa .40-.75) and poor reliabilty (i.e. < .40)?

 

  1. Please fill in Table 2.

Table 2. Factors associated with answering yes to less socially desirable activities on V-CASI in bivariate analysis  

  Unadjusted Odds Ratio (95% C.I.)
Age < 18 vs. >18  
V-CASI first vs. FTFI first  
Comfortable with computers vs. not comfortable  
> 2 sex partners  
Had a new sex partner  
Used a condom last vaginal sex   
Used condoms more than half of the time  
History of gonnorrhea  
Douched  

Please read the following article (available in the Epi department office)

Turner CF; Ku L; Rogers SM; Lindberg LD; Pleck JH; Sonenstein FL. Adolescent sexual behavior, drug use, and violence: increased reporting with computer survey technology. Science. 1998; 280:867-73.
 
  1. Compare our study to the C.F. Turner article. What are the limitations/strengths to each study?


 
 

 

  1. Compare our findings to those of C.F. Turner et al. Were we consistent with what they found? Where there any differences between what we found and what they found?